school
Macau Periodical Index (澳門期刊論文索引)
- Author
- Wan, Yinbo; Liang, Yong; Ding, Liya
- Title
- Mining multilevel association rules from primitive frequent itemsets
- Journal Name
- 澳門科技大學學報
- Pub. Info
- Jun. 2009, Vol.3, No.1, pp. 10-19
- Keyword
- Rule mining；Association rules；Multilevel association rules/frequent itemsets；FP-tree；conept hierarchy
- Abstract
- Abstract : Association rule mining has attracted wide attention in both research and application areas recently. The mining of multilevel association rules is one of the important branches of it. In most studies, multilevel rules will be mined at each individual level of concept hierarchy by scanning the database repeatedly. It affects the efficiency of the mining algorithms, and reduces the integrality and flexibility of the multilevel rules. In this paper, a novel method is proposed to extract multilevel rules based on different hierarchical levels by organizing and extracting frequent itemsets mined from primitive dataitems. Its mining process is based on dynamic concept hierarchies. The performance of the new method is evaluated by restaurant data from real world and the results are promising. Paragraph Headings: 1. Introduction 2. A method for mining multilevel association 3. Mining with dynamic concept hierarchy 4. Performance evaluation 5. Discussion 5.1. Two limitations of our proposed algorithms 5.2. The tradeoff between common sense and specific patterns 6. Conclusion Tables: 1. Comparison of running time (the time value is mean for 10 runs) Figures: 1. An example of concept hierarchy 2. FP(0)-tree (atomic level) 3. Uncompleted FP(1)-tree (step1-2) 4. Uncompleted FP(1)-tree (step3) 5. Uncompleted FP(1)-tree (constructed by algorithm CFP) 6. The mining flow of different concept level